Tenet Security raises $6M for AI agent runtime controls

Enterprise adoption of AI agents is accelerating, but security teams lack visibility into how these autonomous systems interact with production environments. Tenet Security, a new startup founded by former Cisco AI Defense engineers, aims to address this gap with runtime controls that simulate and block risky agent actions before they execute.

The company emerged from stealth on 19 June 2026 with $6 million in seed funding led by The Westly Group and MizMaa Ventures. Its patent-pending Agent-side Simulation technology predicts an agent’s likely next steps and intervenes if the path appears dangerous, generating an audit trail to explain each decision. This approach targets what Tenet calls "Agentjacking"—attacks that manipulate AI agents into performing harmful actions while operating within their authorized permissions.

Runtime security challenges

Traditional security controls focus on user identities, endpoints, and network traffic, but AI agents introduce new risks. An agent with valid credentials might still execute harmful actions if influenced by malicious instructions embedded in emails, logs, or databases. These attacks often evade detection because they don’t rely on conventional malware or unauthorized access.

Tenet’s founders, Barak Sternberg and Nevo Poran, previously built offensive security tools for Fortune 500 companies at Wild Pointer before joining Cisco’s AI Defense team. Their experience informs the company’s focus on attacker behavior rather than compliance. The startup claims its threat lab has validated Agentjacking techniques across over 100 enterprise environments, identifying thousands of organizations with publicly accessible attack paths that existing security tools missed.

Market and adoption hurdles

Early deployments suggest demand for runtime controls. A legal-sector enterprise reportedly expanded its AI agent deployments from two to over 20 in six months while using Tenet’s platform, blocking more than 10 attempts, including a cross-site scripting attack. Another Fortune 1000 customer discovered a misconfigured agent generating tens of thousands of dollars in unnecessary token costs over a single weekend, highlighting how agent risks extend beyond security breaches to operational and financial losses.

However, the market remains nascent. Enterprises are unlikely to standardize on a single agent framework, as these systems are embedded in SaaS tools, cloud platforms, and internal workflows. Tenet’s platform must integrate with multiple frameworks to avoid becoming a niche solution. The company also faces competition from larger security vendors, which may absorb runtime controls into broader platforms.

What to watch

Tenet’s funding will support product development, threat research, and hiring in North America. The company’s success may hinge on its ability to scale coverage across diverse agent frameworks while proving its technology works in complex production environments. Regulatory scrutiny could also accelerate adoption if agencies demand auditability for AI-driven workflows in regulated sectors like finance and healthcare.

The broader question is whether enterprises will prioritize security early in their AI agent rollouts or react only after incidents occur. Tenet’s approach—preventing risky actions rather than detecting them after the fact—could gain traction if it balances security with operational continuity.